POLLING companies failed to predict the result of last year's General Election because they asked the wrong people, a study by Britain's leading psephologist has concluded.

On the eve of the poll last May, almost all the polling companies believed Labour was neck and neck with the Conservatives only for David Cameron's party to achieve a decisive victory.

It was their worst election since 1992 when they forecast Neil Kinnock to become prime minister before his Labour Party was well beaten by John Major's Conservatives.

In a report for NatCen Social Research, Professor John Curtice, of Strathclyde University, said the pollsters did not use representative samples of the public.

A number of theories have been put forward to explain the error, including the idea that "shy Tories" were reluctant to tell pollsters of their true voting intentions.

The failure has also been blamed on "lazy Labour" supporters, who responded to polls but did not bother to vote, or a very late swing to the Conservatives which went undetected.

Dismissing those notions, Professor Curtice concluded the polling companies simply questioned too many Labour supporters and too few Conservatives, and failed to correct the mistake when they came to analyse their data.

He suggested pollsters struggled because Conservative voters were harder to make contact with over the phone or by email - the ways in which most polls are conducted.

With just two or three days to question more than 1000 people from their list of possible contacts, the companies spoke to a disproportionate number of Labour supporters, who were more accessible.

Pollsters also underestimated the likelihood that younger people would not turn out to vote.

The mistakes were not corrected by the "weighting" process, which is designed to adjust raw polling figures to create a representative picture of the population as a whole, Professor Curtice said.

His conclusions were based on comparing opinion poll results with the annual British Social Attitudes Survey.

The survey uses a genuinely random sample of more than 4000 people.

When, in the months after the election, the group was asked how they had voted, their answers reflected the result far more accurately than the opinion polls.

Professor Curtice said: "Much of the speculation about the failure of the polls in the general

election has focused on claims that voters were not being honest

with the pollsters – and perhaps even to themselves.

"However, the British Social Attitudes Survey's relative success in replicating the outcome of the 2015 election suggests that the problem of the polls lay instead in the

unrepresentative character of the samples of people whom they

interviewed."

Most polling companies had Labour and the Conservatives within one or two percentage points of each other on the eve of the election.

In the end, David Cameron's party took

37.8 per cent of the vote and Labour 31.2 per cent, a margin of 6.6 percentage points.

Throughout the campaign, polling suggested Britain was heading for a hung parliament and influenced the way parties campaigned.

The Conservatives tried to capitalise on the possibility of an Ed Miliband government being propped up the SNP, a scenario that was unpopular in England.

After the election industry bodies the British Polling Council and the Market Research Society launched an independent inquiry into why the pollsters got it so wrong, headed by Professor Patrick Sturgis of the University of Southampton.

He is due to present his preliminary findings next week.